Basically it means that BOINC thinks based on what you have left in your cache + how long it will take to crunch them + how much time you have until the deadline + what your "connect every x days" setting is set for, that you might not be able to report that task before the deadline, so it gets moved up to high-priority to make sure it completes before the deadline.

Usually this means the estimates have not sorted themselves out yet, but they will after 10 completed tasks, or it means that for some reason or another, you ended up with way too much work.

---

Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)

And sometimes tasks go HP for no apparent reason whatsoever, especially if you run Einstein as well as Seti.

I'm sure there is a reason, but it must be governed by a very complex formula.

---

David
Sitting on my butt while others boldly go,
Waiting for a message from a small furry creature from Alpha Centauri.

I can artificially generate "high priority" by limiting the number of cores I'm using for SETI (which I do while gaming). Suddenly from the POV of the manager, it sees it'll take three times longer to process all the CPU units assigned which impacts shorties that only have a delivery date like two weeks after being assigned (instead of like six weeks for normal units) so any shorty I have waiting to be processed gets bumped up to the run NOW, while the unit that was being processed gets put on standby.

---

"Life is just nature's way of keeping meat fresh." - The Doctor

It is also worth noting that it will depend on your BOINC version, which task will run in HP. Some versions strictly enforce EDF - earliest deadline first - while others will choose tasks further down the list - those that the simulation has decided will not make it in time if the normal FIFO (first in first out) order of running is used.
That can sometimes cause confusion.

[for the record, 6.10.x and 7.0.14 and later do strict EDF, 6.12 runs 'bottom up']

As to reason for HP - there is a very complex simulation running at all times, about when tasks will finish, based on the runtime estimate.
Whenever the runtime estimates go bad, the simulation get fed wrong assumptions and comes to wrong conclusions.

Another scenario that habitually sees tasks in HP is machines that are not constantly running.
Let's say you have BOINC running (legally, i.e. you asked your boss) on your work computer. That computer though gets turned off in the evening and for the weekend. BOINC keeps an longterm avarage of the time it is running to tweak cache levels, so it will not in fact ask for too much work. But invariably, it will panic on Monday morning, when that average dips due to the weekend shutdown.

Just remember 'em big friendly letters. It is very rare for BOINC to actually miss deadlines under normal operation. It's not BOINC's fault when you go on holidays and forget to run down your cache ;)

---

A person who won't read has no advantage over one who can't read. (Mark Twain)

Thanks. I just looked at the deadline dates on the stuff sent to me, and while most is due April and into May, these two are due 3/26 and were buried down the list.

If you Click on The Little Arrow in the Heading of the Column 'Estimated Time remaining'it will Rearrand Your List. So it may not have actually been 'Buried' in The List, it may have just been the way it was displayed.

---

GALAXY-VOYAGER

Basically it means that BOINC thinks based on what you have left in your cache + how long it will take to crunch them + how much time you have until the deadline + what your "connect every x days" setting is set for, that you might not be able to report that task before the deadline, so it gets moved up to high-priority to make sure it completes before the deadline.

Usually this means the estimates have not sorted themselves out yet, but they will after 10 completed tasks, or it means that for some reason or another, you ended up with way too much work.

What about when freshly *Downloaded Tasks are Added at the Top of The List above Currently Running Tasks, which are suspended upon Completion of such Download, and the New One is Started and Run as HP. And when *it finishes, another fresh downloaded task does the same thing. Hence resulting in the remaining Tasks being given less chance of Completing On Time ?

---

GALAXY-VOYAGER

Thanks. I just looked at the deadline dates on the stuff sent to me, and while most is due April and into May, these two are due 3/26 and were buried down the list.

If you Click on The Little Arrow in the Heading of the Column 'Estimated Time remaining'it will Rearrand Your List. So it may not have actually been 'Buried' in The List, it may have just been the way it was displayed.

And after you've done that you might want to check this post for how to get the boinc manager to display the tasks in order of arrival again...

---

A person who won't read has no advantage over one who can't read. (Mark Twain)

Basically it means that BOINC thinks based on what you have left in your cache + how long it will take to crunch them + how much time you have until the deadline + what your "connect every x days" setting is set for, that you might not be able to report that task before the deadline, so it gets moved up to high-priority to make sure it completes before the deadline.

Usually this means the estimates have not sorted themselves out yet, but they will after 10 completed tasks, or it means that for some reason or another, you ended up with way too much work.

What about when freshly *Downloaded Tasks are Added at the Top of The List above Currently Running Tasks, which are suspended upon Completion of such Download, and the New One is Started and Run as HP. And when *it finishes, another fresh downloaded task does the same thing. Hence resulting in the remaining Tasks being given less chance of Completing On Time ?

A project that is running in HP should NOT fetch.
That may be different in recent versions, where workfetch is governed by the amount of work recently done for a project, besides the cache levels.

However if you throw the system out of balance (and starting fresh IS out of balance) it can take a while to settle down.

---

A person who won't read has no advantage over one who can't read. (Mark Twain)

Thanks. I just looked at the deadline dates on the stuff sent to me, and while most is due April and into May, these two are due 3/26 and were buried down the list.

If you Click on The Little Arrow in the Heading of the Column 'DEADLINE'it will Rearrange Your List. So it may not have actually been 'Buried' in The List, it may have just been the way it was displayed.

(I just Edited the above information I Posted earlier: I previously stated the Column Labelled 'Estimated Remaining Time'.... however, it should be 'Deadline' as I have Edited it to. Sorry if I caused any confusion by Posting the incorrect information before).

---

GALAXY-VOYAGER

Basically it means that BOINC thinks based on what you have left in your cache + how long it will take to crunch them + how much time you have until the deadline + what your "connect every x days" setting is set for, that you might not be able to report that task before the deadline, so it gets moved up to high-priority to make sure it completes before the deadline.

Usually this means the estimates have not sorted themselves out yet, but they will after 10 completed tasks, or it means that for some reason or another, you ended up with way too much work.

What about when freshly *Downloaded Tasks are Added at the Top of The List above Currently Running Tasks, which are suspended upon Completion of such Download, and the New One is Started and Run as HP. And when *it finishes, another fresh downloaded task does the same thing. Hence resulting in the remaining Tasks being given less chance of Completing On Time ?

If Normal tasks keep getting pre-empted in favour of shorter deadline tasks, then eventually those Normal tasks will get into deadline trouble,
then Boinc will do those in High priority, and will likely stop asking for work from that project until that hurdle is overcome,
(depends on the number of cores available, and the split of work across them)

Claggy

Basically it means that BOINC thinks based on what you have left in your cache + how long it will take to crunch them + how much time you have until the deadline + what your "connect every x days" setting is set for, that you might not be able to report that task before the deadline, so it gets moved up to high-priority to make sure it completes before the deadline.

Usually this means the estimates have not sorted themselves out yet, but they will after 10 completed tasks, or it means that for some reason or another, you ended up with way too much work.

What about when freshly *Downloaded Tasks are Added at the Top of The List above Currently Running Tasks, which are suspended upon Completion of such Download, and the New One is Started and Run as HP. And when *it finishes, another fresh downloaded task does the same thing. Hence resulting in the remaining Tasks being given less chance of Completing On Time ?

If Normal tasks keep getting pre-empted in favour of shorter deadline tasks, then eventually those Normal tasks will get into deadline trouble,
then Boinc will do those in High priority, and will likely stop asking for work from that project until that hurdle is overcome,
(depends on the number of cores available, and the split of work across them)

Claggy

Okay, thanks for the information. The Computer I'm referring to is a HP Notebook G64 with Windows 7. It's Dual Processor and I presently only have my preferences set to use 1 (one) Processor at 100%. I'm not using both processors because I only have 2GB RAM and I use the machine for General Work whilst SETI is running (and CPU Usage is between 60% and 80% at the Best of times: But often runs at 70%-99%).
I ONLY run SETI on this machine. There are currently about 69 tasks in The Queue (including 2 partially completed ones) It only runs One Task at any one time.
What I was thinking of trying, was to select 'No New tasks' until I cleared some of the Existing ones. do you think that would this be worth Trying.
I mostly have it running 24/7, except i may put it in Sleep mode for a few hours each couple of days, and once or twice a week I Shutdown for 4 to 6 hours or overnight (just to give the system a Rest).
Could I possibly Reset Preferences to use both Processors without causing problems? If so, should I only run it for a Certain Number of hours per day?

---

GALAXY-VOYAGER

Basically it means that BOINC thinks based on what you have left in your cache + how long it will take to crunch them + how much time you have until the deadline + what your "connect every x days" setting is set for, that you might not be able to report that task before the deadline, so it gets moved up to high-priority to make sure it completes before the deadline.

Usually this means the estimates have not sorted themselves out yet, but they will after 10 completed tasks, or it means that for some reason or another, you ended up with way too much work.

What about when freshly *Downloaded Tasks are Added at the Top of The List above Currently Running Tasks, which are suspended upon Completion of such Download, and the New One is Started and Run as HP. And when *it finishes, another fresh downloaded task does the same thing. Hence resulting in the remaining Tasks being given less chance of Completing On Time ?

A project that is running in HP should NOT fetch.
That may be different in recent versions, where workfetch is governed by the amount of work recently done for a project, besides the cache levels.

However if you throw the system out of balance (and starting fresh IS out of balance) it can take a while to settle down.

okay thanks. however, what is actually happening is when The HP has completed, it then fetches New Work. Or, if it does not Immediately fetch new work, and resorts back to one that it was previously processing, when it Uploads THAT one alone, or with any additional ones it has since completed, it then fetches new Work, which recently has been the HP Ones. It not actually Fecthing new Work whilst a HP is in Progress.

---

GALAXY-VOYAGER

Basically it means that BOINC thinks based on what you have left in your cache + how long it will take to crunch them + how much time you have until the deadline + what your "connect every x days" setting is set for, that you might not be able to report that task before the deadline, so it gets moved up to high-priority to make sure it completes before the deadline.

Usually this means the estimates have not sorted themselves out yet, but they will after 10 completed tasks, or it means that for some reason or another, you ended up with way too much work.

What about when freshly *Downloaded Tasks are Added at the Top of The List above Currently Running Tasks, which are suspended upon Completion of such Download, and the New One is Started and Run as HP. And when *it finishes, another fresh downloaded task does the same thing. Hence resulting in the remaining Tasks being given less chance of Completing On Time ?

If Normal tasks keep getting pre-empted in favour of shorter deadline tasks, then eventually those Normal tasks will get into deadline trouble,
then Boinc will do those in High priority, and will likely stop asking for work from that project until that hurdle is overcome,
(depends on the number of cores available, and the split of work across them)

Claggy

Okay, thanks for the information. The Computer I'm referring to is a HP Notebook G64 with Windows 7. It's Dual Processor and I presently only have my preferences set to use 1 (one) Processor at 100%. I'm not using both processors because I only have 2GB RAM and I use the machine for General Work whilst SETI is running (and CPU Usage is between 60% and 80% at the Best of times: But often runs at 70%-99%).
I ONLY run SETI on this machine. There are currently about 69 tasks in The Queue (including 2 partially completed ones) It only runs One Task at any one time.
What I was thinking of trying, was to select 'No New tasks' until I cleared some of the Existing ones. do you think that would this be worth Trying.
I mostly have it running 24/7, except i may put it in Sleep mode for a few hours each couple of days, and once or twice a week I Shutdown for 4 to 6 hours or overnight (just to give the system a Rest).
Could I possibly Reset Preferences to use both Processors without causing problems? If so, should I only run it for a Certain Number of hours per day?

Just don't cache so much, and don't worry about it. High priority is designed to prevent tasks over-running their deadlines - it's perfectly normal and doesn't affect your computer at all.

Okay, thanks for the information. The Computer I'm referring to is a HP Notebook G64 with Windows 7. It's Dual Processor and I presently only have my preferences set to use 1 (one) Processor at 100%. I'm not using both processors because I only have 2GB RAM and I use the machine for General Work whilst SETI is running (and CPU Usage is between 60% and 80% at the Best of times: But often runs at 70%-99%).
I ONLY run SETI on this machine. There are currently about 69 tasks in The Queue (including 2 partially completed ones) It only runs One Task at any one time.
What I was thinking of trying, was to select 'No New tasks' until I cleared some of the Existing ones. do you think that would this be worth Trying.
I mostly have it running 24/7, except i may put it in Sleep mode for a few hours each couple of days, and once or twice a week I Shutdown for 4 to 6 hours or overnight (just to give the system a Rest).
Could I possibly Reset Preferences to use both Processors without causing problems? If so, should I only run it for a Certain Number of hours per day?

What cache settings are you running?

Claggy

As others have said, if you leave BOINC alone, it will sort itself out eventually.

Probably the reason why freshly downloaded tasks immediately go into HP mode is probably because the estimate for them is way high.

For example, I only just last week allowed my slow single-core machine to do AP_v6 for the first time since it came out. It managed to snag a WU and it was given an estimate of 384 hours. The actual number (with r557) comes out to about 49 hours. I have only done two APs on that machine so far (I let it grab whatever is available when it asks for work..I don't do only one or the other), so the estimates are still going to be nearly 400 hours until I have completed 10 APs. And by completed tasks, that means less than 10% blanked, and did not early-exit with 30/30, so there's always the chance it will take more than 10 actual APs to do that.

---

Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)

i could be wrong here (really it does happen now and then) but i think that if a task have been resent many times with no quorum then when it gets sent out as a new wu download it gets set to high priority.

---

I RTFM and it was WYSIWYG then i found out it was a PEBKAC error

Re-sends are sent out with a normal length deadline. And so run a normal priority, unless the rules outlined in earlier posts apply.

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

We may have an issue with high priority. I have every task that is running is running under high priority. Two of them are not due until 04/02 and my computer runs 24 hours a day and can handle 4 tasks at once. One for 04/02 is waiting and then until 04/08 all are under 2 hours estimated time. I don't think these should be high priority as they will all be finished by 03/22/13 which is 13 days before they are due.

Ian Fraigun

We may have an issue with high priority. I have every task that is running is running under high priority. Two of them are not due until 04/02 and my computer runs 24 hours a day and can handle 4 tasks at once. One for 04/02 is waiting and then until 04/08 all are under 2 hours estimated time. I don't think these should be high priority as they will all be finished by 03/22/13 which is 13 days before they are due.

Ian Fraigun

What cache settings are you running? If you're running a 10+0 days cache, then also saying you're going to be offline for 10 days, so any Wu's need to be reported 10 days early, try reducing your cache settings.

Claggy

There's a weird thing going on. Seen it with other people. There are a lot of tasks out there at this time that have, e.g.

Created 20 Mar 2013, 15:31:12 UTC
Sent 20 Mar 2013, 18:40:29 UTC
Report deadline 20 Mar 2013, 22:45:19 UTC

Or

Created 20 Mar 2013, 10:57:53 UTC
Sent 20 Mar 2013, 15:58:02 UTC
Report deadline 20 Mar 2013, 16:45:04 UTC

And that they go to GPU only, I can understand... but no, they go to CPU as well.
Look at this guy, he had 60 of them, going to his CPU and GPU! Even if that's an i7 with 8 threads and a GPU, it's a bit much to run 60 tasks in the allotted 6 and 3/4 hours.

---